Agentic AI Tech Lead

I lead agentic AI adoption for the NFG engineering team. I turn Claude Code from a coding assistant into an orchestrated delivery system built on custom skills, subagents and MCP integrations.

• End-to-end ticket orchestration: I built skills that chain a Jira ticket into an isolated git worktree, then implementation, a PR with the right reviewers, CI, a shared UAT/Beta merge, verification and the Jira transition.
• Multi-agent coordination: I run parallel agents and route each task to the right model. Cheaper models gather data and references, and stronger ones handle design and code. The work spans 15+ repos and worktrees.
• Process as code: I turned team working agreements into reusable skills: PR review SLAs, Jira status definitions, ticket templates and pair-testing plans. The process is enforced the same way every time.
• AI-driven security remediation: I handled pen-test findings (file-upload, IDOR, XSS, session timeout) and CVE research. Each fix comes with dependency diffs, Snyk baselines and fix manifests.
• Persistent agent memory: I documented environment quirks, auth behavior and tooling pitfalls so agents keep context across sessions and don’t repeat mistakes.
• CI/CD automation: I’m automating GitHub Actions versioning and package publishing across about 11 internal Ruby gems.
• Stack: Claude Code, MCP (Atlassian), skills/subagents, Rails, React, Auth0, GitHub Actions, Snyk.

Agentic AI Tech Lead

LIV is a multi-tenant platform for clinical intake and psychiatric screening, in Hebrew and English. It spans 9 repos and about 750k lines: NestJS/Prisma, React 19, FastAPI LLM services on Azure OpenAI, safety and anonymization services, and Terraform/Helm on Azure and GCP. About 30 engineers contributed.

• Agent workflow design: Defined a shared, model-agnostic skill library that covers a ticket from start to finish:
  • Create Clarity: checks the issue against the real code and turns it into a ticket backed by evidence.
  • Implement: takes one issue to one open PR, inside its own isolated git worktree.
  • Watch-and-Fix / Watch-and-Review: a developer agent and a reviewer agent run against the same PR at the same time, often on different models (Claude as developer, Codex as reviewer). They coordinate only through PR comments that carry identity markers.
  • Manual QA and Clean Memory: QA in an isolated local Kubernetes environment, and cleanup of leftover worktrees and resources.
• Running agents in parallel safely: Each agent run had its own worktree, guarded by run locks and per-target locks. I had more than 30 worktrees active on some days. A script picked the right GitHub account for each agent's role, so several team members and their agents could share the repos without collisions.
• Human control: Agents recommend the next step but never start it on their own. Humans merge every PR. Architecture proposals and schema changes need approval before any code is written.
• Delivery: 233 commits between July 21 and August 5, 2026, mostly in liv1.0 and the new Clinician Editor. Highlights:
  • Set up the Clinician Editor web shell, its CI and CODEOWNERS.
  • Added a tenant-neutrality scanner with a required repository guard.
  • Added a check that a stack of dependent PRs was actually delivered, verified by containment rather than by timestamps.
  • Drove chat embodiment (the avatar and voice) from the tenant projection.

Agentic AI & Fullstack Tech Lead

• AI Skills improvement for optimized developer experience and code alignments
• Build from scratch mobile-web application UI and fix responsive web UI
• Architected and maintained Auth0 Identity-as-Code across Dev, Staging, and Production tenants using Terraform + Terragrunt, managed via GitHub Actions CI/CD pipelines
• Implemented enterprise SSO integrations with Microsoft Azure AD (WAAD) and Okta, including Home Realm Discovery (HRD) and identifier-first login flows
• Built full-stack features end-to-end - from OpenSearch index schema, DynamoDB data models, and AWS Chalice (Lambda) REST endpoints to React + TypeScript UI components
• Developed a Project Scope feature across the entire stack: API design, feature flags, Jotai state atoms, custom hooks, and settings UI tabs
• Implemented feature flag infrastructure (SettingsFeature, ClientFeatureFlags, FeatureFlagResolver) powering gradual rollout across client and server
• Hardened the authentication session lifecycle - migrated from popup-based to redirect-based OAuth flow with improved error handling
• Maintained and extended the Python/Chalice microservices monorepo spanning gateway, onboarding, identity, media processing, and enrichment services
• Managed an Auth0 users allowlist system with integration to Cognito fallback for user-existence checks
• Set up GitHub Actions workflows for automated hotfix syncing between release and main branches
• Exposed FastAPI auto-docs at a public endpoint for internal developer tooling
• Tracked and instrumented BI analytics events wired into the product analytics pipeline
• Worked with Turborepo monorepo tooling, Playwright E2E tests, and Jest unit tests across a multi-app frontend workspace
• Contributed to AWS serverless infrastructure across Lambda, DynamoDB, OpenSearch, and Cognito in a multi-environment setup

Agentic AI & FullStack Tech Lead

Built a platform that produces verified identity reports for government and insurance clients, combining AI-driven document analysis with domain-expert review.

• Led development of the Next.js application (frontend and API routes), deployed across AWS, GCP and Azure.
• Built large-scale Playwright scrapers that crawl government and public-records websites, handle pagination, dynamic content and session flows, and download public legal documents automatically.
• Designed a distributed ingestion pipeline on AWS SQS and SNS that queued, retried and fanned out millions of PDFs and scanned images to processing workers.
• Built an LLM document-processing pipeline that scans, classifies and extracts structured data (names, dates, case numbers, parties, rulings) from unstructured legal PDFs and images.
• Built a multi-provider LLM layer over Gemini, OpenAI and self-hosted Ollama. Each task was routed by cost, accuracy and data sensitivity: local models handled high-volume or sensitive documents, and cloud models handled complex reasoning.
• Used Gemini's multimodal vision to read scanned and handwritten documents and images directly, cutting the need for a separate OCR step.
• Engineered prompts and schema-validated JSON output so LLM results were consistent, machine-readable and safe to store, with validation and fallback when output was malformed.
• Generated concise LLM summaries of long legal filings, which cut analysts' manual review time when building verification reports.
• Modeled verification data in PostgreSQL with Prisma, linking source documents, extracted entities and report results for full traceability.
• Implemented NextAuth with Google Sign-In, role-based user management and secured API access.
• Wrote end-to-end Playwright tests to keep critical verification and reporting flows reliable.

FullStack Tech Lead

Working on Node.js Server Application for Time Reporting

• ⁠ Replace an existing PHP web application with a Node.js-based server application.
• Integrate with Google BigQuery for data storage and analysis.
• Support a Flutter mobile application for user interaction.

Mobile Architect & Team Leader

Joined the company in order to build from scratch their mobile application using Flutter with highly security protocols.

The company has developed a platform to manage and controller the users crypto assets in a single place while keeping a self custodian approach. Only the user is responsible for his assets.

Providing both Web & Mobile application integrating with secured backend to manage transactions, coin swap and user management roles and more inside an organization.

• Mobile Architect & Team Leader
• Building Serverless application using NodeJs & Express on GCP for Remote Messaging (Notifications), remote configuration and reactive data base events on Firestore
• Architecture & Development of Flutter application, using uni-directional data flow in a multi-module application with Repository Pattern.
• Building multi step authentication for login, biometric access and backend states.
• Building a secured cryptography signing using device CPU to verify approvals or rejections of transactions or user management to avoid frauds.
• Using Crashlytics and BigQuery for analytics and error reporting.
• Using Firebase Authentication for personalized experience.
• Using GetX state management & navigation, GetIt (Dependency Injection)
• Custom Themes & UI
• Multi Flavor application support Dev, Staging, Prod for both Android and iOS in stores.
• Job Interviews for Flutter Senior Developers to join the team.

Mobile Architect

Joined the team to build from scratch and set the guidelines on Flutter mobile architecture & development for a new matching experience application using an AI bot.

This startup has a revolutionary way to attempt and create the perfect match for the user by understanding and getting the know the user.

Suggestions are based on preferences and dynamic content made by psychologists and data fed by the user.

• Mobile Team Leader
• Architecture & Development of Flutter application, using uni-directional data flow in a multi-module application with Repository Pattern.
• Build dynamic bot content using AWS Amplify as a backend
  (Cognito Authentication, S3 storage) with MongoDB for data input to feed AI service.
• Build Admin application/web admin for future change to dynamic content
• Support dynamic UI changes
• Flutter Web Support
• Using GetX state management, GoRouter2 navigation, GetIt (Dependency Injection)
• Custom Themes & UI
• Job Interviews for Flutter developers to join the team

Consultant - Mobile Architect

Joined the team to create from scratch and give correct guidelines on Flutter mobile architecture & development for solar plate installers.

The project was started in order to replace the existing application that was done as a POC in the company and was then decided to take it to the next level.

• Architecture & Development of Flutter application, using uni-directional data flow in multi-module application with Repository Pattern.
• Flutter Web Support
• Using GetX state management, GoRouter2 navigation, GetIt (Dependency Injection), dynamic localization/translations, Dio with SSO
• Custom Themes

Consultant - Mobile Architect

Joined the team to build MyOfer Malls mobile application architecture. 

The project was started in order to replace the previous application that was in the stores, with many mall-related features and user personalization.

• Architecture & Development of Flutter application, using uni-directional data flow in multi-module application with Repository Pattern.
• Using BLoC state management, GetIt (Dependency Injection), dynamic localization/translations, Nested Navigation with complex UI

Mobile Expert

The project is all about making a simple infusion pump SMART. 

Once the Patient will have the pump at home and once he gets his treatment the mobile application will connect from the background and pull the data (events) and report them to the Doctor or his CareGiver.

• Architecture & Development for Mobile application to integrate with Infusion Pump via BLE.
• Architecture & Development for Cloud Solutions using Serverless Framework on AWS & Firebase
• Building Flutter UI application, multi-module with repository pattern.
  Implementing BLoC State Management, GetX, Localization, Dio, Encryption, GetIt Dependency Injection, and more
• Developing Native Code (Kotlin, Java & Swift) for BLE communication and background services (WorkManagers, Receivers, Services)
• Azure Pipeline Dev-ops for building & deploy automation
• Support HIPPA and GDPR regulations (Encrypting Data-at-rest & Data-in-transit)

Mobile Expert

• The project is all about making a Simple infusion pump Smart.
• Once Patient will have the pump at home and once he gets his treatment the mobile application will connect from the background and pull the data (events) and report them to the Doctor or his CareGiver without the patient needing to go with the actual pump to the pharmacy in order to update the care giver with his treatment status and progress.
• Architecture & Development for Mobile application to integrate with Infusion Pump via BLE.
• Architecture & Development for Cloud Solution using ServerLess Framework on AWS Building Flutter UI application, Implementing BLoC StateManagement, GetX, Localization, Dio, RxDart, Encryption and more
• Developing Native Code (Kotlin, Java & Swift) for BLE communication and background services (WorkManagers, Receivers, Services, FCM)
• Support HIPPA and GDPR regulations (Encrypting Data-at-rest & Data-in-transit)

Mobile Team Leader & Senior Android Developer

I Am Not Alone is a growing startup in the health sector that created a patented, highly secured data collection and patient management platform that is disorder agnostic.

- Managing development teams in the company. (Backend, Frontend & Mobile)

- Architecture and Development of Android Applications from Scratch using MVP, MVVM, and Repository Pattern.

- Support HIPPA and GDPR regulations (Encrypting Data-at-rest & Data-in-transit)

- Architecture of entire medicine, measurements of patient follow-ups in order to collect and present to caregivers.

- Integration to Treatment plans from Hospitals and Medical institutions across the globe.

Senior Android Developer

Vi develops products to help people exercise through personalized, goal-oriented coaching and social play.

I joined the team to refactor the entire UI/UX after the product have changed.

- Working with MVVM & Dagger2 (dependency injection)

- Connecting Media Player to the application.

- Working with Broadcast Receivers and Event Driven Application architecture in order to update UI from Vi AI SDK

Senior Android Developer

OpenWeb turns publishers into hosts, integrating into a third-party application or website in order to add User sign-in, chat communication, and more...

Developed Mobile SDK and Sample project to implement and test the SDK.

- Building everything from scratch in order to make sure SDK is as slim as possible.

- Designing and adding new features for the SDK

- Helping clients with the integration process

- Solving bugs related to specific devices and client integration.